Euclid Catalogue of Galaxy Clusters (ECGC)

• ECGC is a validated, multi-tier catalogue that integrates Euclid-native optical detections with external multi-wavelength counterparts across 14,000 deg².
• It employs dual detection pipelines—AMICO and PZWav—to deliver robust, high-purity cluster identifications, serving as a precursor to the final catalogue.
• The layered catalogue design supports both astrophysical studies and cosmological analyses by offering observables, mass-calibration data, and covariance-aware abundance statistics.

The Euclid Catalogue of Galaxy Clusters (ECGC) is the Euclid mission’s cluster-catalogue programme: a large, optically selected survey product expected over approximately $14{,}000$ square degrees of extragalactic sky, extending well beyond redshift unity and intended for both astrophysical and cosmological analyses. The published literature presents the ECGC not as a single undifferentiated list, but as a validated, provenance-rich catalogue ecosystem in which Euclid-native detections, external multi-wavelength counterparts, mass-calibration layers, and specialized subcatalogues are combined for different science cases. At present, the available Euclid Quick Release 1 (Q1) papers are best understood as precursor products and workflow demonstrations rather than the final ECGC itself (Collaboration et al., 8 Sep 2025, Collaboration et al., 24 Mar 2025).

1. Concept, scope, and survey context

Within Euclid, the ECGC is one of the mission’s flagship survey products. The consortium documents describe it as a catalogue that will contain many new high-redshift clusters, while at lower redshifts a fraction of the systems will already have counterparts in other surveys. The catalogue is intended to support both astrophysical applications—such as galaxy evolution, environment, and structure growth—and cosmological applications based on cluster abundance and clustering. The operational cluster detection in the photometric survey is built around two pipeline algorithms adapted for the Science Ground Segment, AMICO and PZWav, and the ECGC is explicitly not expected to be used as a single monolithic list: “well characterised sub-catalogues” are intended to feed specific analyses, especially cluster cosmology (Collaboration et al., 8 Sep 2025).

The current Euclid on-sky reference point is Q1, the first survey data release over the future Euclid Deep Field locations but at Euclid Wide Survey depth. Q1 covers about $63\,\mathrm{deg}^2$ in three non-contiguous fields—roughly $23\,\mathrm{deg}^2$ in EDF-N, $28\,\mathrm{deg}^2$ in EDF-S, and $12\,\mathrm{deg}^2$ in EDF-F—and functions as a validation ground ahead of DR1, which is expected to cover about $1900\,\mathrm{deg}^2$. The Q1 cluster papers therefore occupy a transitional role: they demonstrate that the Euclid cluster workflow functions on real data, but they do not yet deliver the full ECGC selection function or its definitive release schema (Collaboration et al., 24 Mar 2025).

This staging is methodologically important. The published ECGC-related papers distinguish between at least four layers: Euclid-native optical detections; external validation against X-ray, SZ, weak-lensing, velocity-dispersion, and historical optical catalogues; mass calibration, especially through weak lensing; and specialized high-value subsets such as strong-lensing cluster fields. This suggests a layered ECGC architecture rather than a single homogeneous table.

2. Detection workflow and the Q1 precursor catalogue

The first direct precursor to the ECGC is the Q1 joint cluster catalogue of 426 high-S/N detections in the range $0.2 \le p \le 1.5$, where $p$ denotes the Euclid photometric cluster redshift. These systems were detected independently by both AMICO and PZWav and then combined into a conservative overlap sample designed for purity rather than completeness. The field-by-field split is 137 in EDF-N, 98 in EDF-F, and 191 in EDF-S. After external matching and broader literature checks, 349 of these systems have at least one known counterpart and 77 remain potentially new; the paper further notes that about 20 of the 426 detections are likely spurious after visual and ancillary inspection (Collaboration et al., 24 Mar 2025).

The two detection algorithms play complementary roles. AMICO, the Adaptive Matched Identifier of Clustered Objects, operates directly on galaxy catalogues using angular positions, magnitudes, and full photometric-redshift probability distributions. Its cluster model adopts a Schechter luminosity function and a projected NFW radial profile, while the field component is estimated empirically. PZWav is more model-independent: it searches for overdensities in a data cube of sky position plus redshift using a wavelet-like difference-of-Gaussians kernel with inner and outer Gaussian widths of 300 kpc and 2 Mpc, respectively, and a search redshift bin size of 0.06. In the Q1 implementation, the two catalogues are cross-matched with a physical distance criterion of 1 Mpc converted to angular scale at the mean pair redshift, together with a redshift consistency cut

$\delta z = 0.1(1+z).$

This overlap strategy is already a concrete precursor of ECGC quality control (Collaboration et al., 24 Mar 2025).

Q1 also fixed several pragmatic boundaries that are informative for later ECGC releases. The main sample was restricted to $0.2 \le p \le 1.5$ because of the quality of the early Euclid photometric redshifts, the current exclusion of NISP-only detections owing to persistence issues, and residual photometric-redshift bias and PDZ-width miscalibration. The galaxy input catalogue was limited to $63\,\mathrm{deg}^2$0 objects with $63\,\mathrm{deg}^2$1, appropriate band coverage, star rejection through $63\,\mathrm{deg}^2$2, and additional masking around bright stars and artifacts. These are Q1-specific operational choices rather than final ECGC rules, but they show how source-level quality cuts propagate into the cluster product (Collaboration et al., 24 Mar 2025).

The Q1 paper also explored an extension to $63\,\mathrm{deg}^2$3, finding 58 joint detections in that interval and retaining 15 as the “most reliable” high-$63\,\mathrm{deg}^2$4 candidates after visual vetting. This high-$63\,\mathrm{deg}^2$5 tail was explicitly kept outside the validated main sample, which underscores that the Q1 catalogue is a precursor defined by reliability thresholds rather than a complete ECGC-like census (Collaboration et al., 24 Mar 2025).

3. External validation and multi-wavelength counterparting

External validation is treated as a constitutive part of ECGC construction rather than a post hoc annotation step. The validation methodology paper gives four explicit reasons for this layer: confirming that Euclid detections are bona fide clusters, identifying newly discovered clusters, preparing scaling-relation analyses between Euclid and external observables, and characterizing the Euclid cluster selection function. The authors argue that simple fixed-aperture sky matching is insufficient because optical detections trace galaxy overdensity rather than the hot intracluster medium or the deepest potential minimum, so the matching must instead be based on physical consistency in redshift and scale, with visual inspection for complex cases (Collaboration et al., 8 Sep 2025).

To support this, the consortium assembled a suite of catalogues and meta-catalogues. On the X-ray and SZ side, these include MCXC-II, MCSZ, ComPRASS, the M2C master table, and an internally constructed eROSITA compilation. On the optical support side, they include the LC$63\,\mathrm{deg}^2$6 weak-lensing meta-catalogue, a Global Abell catalogue with 5250 entries, the MCCD velocity-dispersion catalogue with 1082/1083 clusters, and an Optical master table with 6367 single entries. In the demonstration using DES Y1 RedMaPPer as a surrogate for the future ECGC, the updated EC-RedMaPPer product links 1040 out of 6729 optical detections to at least one external counterpart, thereby illustrating how a Euclid-like optical catalogue can be augmented with multi-wavelength information (Collaboration et al., 8 Sep 2025).

The matching logic is deliberately multi-stage. When an external catalogue has a mass estimate, two physical matching methods are used. The first is a mutual-nearest-neighbour “two-way” match in angle, subsequently filtered by normalized separation and a redshift consistency condition

$63\,\mathrm{deg}^2$7

with $63\,\mathrm{deg}^2$8 in the RedMaPPer surrogate demonstration. The second is an $63\,\mathrm{deg}^2$9 search with $23\,\mathrm{deg}^2$0, intended to expose substructures or infalling groups that lie within several characteristic radii of the same halo. When no mass is available, a positional two-way match with catalogue-specific angular thresholds is used instead. Master tables are then used for “loop closure” consistency checks, and visual inspection resolves ambiguous mergers, nearby same-redshift structures, or masking-induced failures (Collaboration et al., 8 Sep 2025).

A complementary external-validation channel is the Euclid adaptation of MCMF for SZ- and X-ray-selected clusters. In Q1 this was applied to overlaps with RASS-MCMF, eROSITA eRASS1, SPT-SZ MCMF, and ACT-DR5 MCMF. The central conclusion is that Euclid can confirm clusters at $23\,\mathrm{deg}^2$1 as effectively as current optical surveys at $23\,\mathrm{deg}^2$2. A particularly practical result is that, for ACT-like random lines of sight at $23\,\mathrm{deg}^2$3, a richness threshold near $23\,\mathrm{deg}^2$4 excludes about 80% of chance superpositions, whereas WISE-based MCMF required $23\,\mathrm{deg}^2$5. This establishes Euclid not only as a source of Euclid-native detections, but also as a high-$23\,\mathrm{deg}^2$6 confirmation and characterization engine for externally selected ICM samples (Klein et al., 24 Jun 2025).

4. Catalogue observables, richness, and mass calibration

The precursor Q1 catalogue already shows the types of observables that will populate the ECGC core. Its joint cluster table contains Euclid cluster name and internal identifier, PZWav and AMICO positions, PZWav and AMICO redshifts, the two finder S/N values, and a photometric-membership richness estimate denoted $23\,\mathrm{deg}^2$7. In the appendix tables, spectroscopic redshift and the number of spectroscopic members within 2 Mpc are also included when available. These are not yet a full ECGC schema, but they define a concrete baseline: positions, redshift, multi-finder provenance, and an internal richness observable (Collaboration et al., 24 Mar 2025).

The Q1 richness measure comes from the LE3 function RICH-CL and is computed from photometric membership probabilities following Castignani & Benoist (2016). Membership probabilities depend on projected cluster-centric distance, $23\,\mathrm{deg}^2$8-band magnitude, and photometric redshift, and the richness is the sum over galaxies brighter than

$23\,\mathrm{deg}^2$9

Empirically, the Q1 paper shows positive correlations between $28\,\mathrm{deg}^2$0 and DES Y1 redMaPPer richness, eROSITA $28\,\mathrm{deg}^2$1, eROSITA $28\,\mathrm{deg}^2$2, and stacked Planck SZ flux. The same paper therefore positions Euclid richness as a meaningful, though early-release, mass-proxy observable (Collaboration et al., 24 Mar 2025).

At the same time, the ICM-overlap study stresses that early-release richnesses need quality metadata. In Q1, richness estimates at $28\,\mathrm{deg}^2$3 are currently limited by photometric issues: the red-sequence colour scatter is larger than in Legacy Survey DR10, and compact bright galaxies with $28\,\mathrm{deg}^2$4 can have masked cores and colour offsets of order $28\,\mathrm{deg}^2$5 mag in at least one colour. The authors therefore treat Q1 richness as a useful but release-dependent quantity and expect substantial improvement in DR1 and later (Klein et al., 24 Jun 2025).

Weak-lensing mass calibration is the most important downstream quantitative layer. The Euclid preparation paper on COMB-CL states explicitly that this combined cluster and weak-lensing pipeline “will measure cluster WL shear profiles and masses for the Euclid Survey.” In the unified precursor analysis across CFHTLenS, DES SV1, HSC-SSP S16a, KiDS DR4, and RCSLenS, the stacked cluster-lensing observable is written as

$28\,\mathrm{deg}^2$6

Using a common pipeline and common modelling assumptions, the study finds agreement between independent surveys “at the level of systematic noise in Stage-III surveys or precursors,” with overlap-region mass accuracy of about $28\,\mathrm{deg}^2$7 and agreement with literature weak-lensing masses at about $28\,\mathrm{deg}^2$8. For Euclid itself, the forecast is that the mission will be able to measure the weak-lensing masses of around 13000 massive clusters with $28\,\mathrm{deg}^2$9 if only shot noise is considered, or around 3000 when noise from shape and large-scale structure is included (Collaboration et al., 2024).

A direct implication is that individual weak-lensing masses will exist only for a minority of ECGC entries. The same COMB-CL study emphasizes that most detected groups and clusters will still require stacked weak-lensing calibration or mass proxies. This suggests that the mature ECGC will need explicit distinctions between detection observables, proxy-based masses, and individual weak-lensing mass products.

5. Statistical interpretation for abundance cosmology

Once the ECGC exists as a large photometric cluster sample, its cosmological exploitation requires a covariance model that goes beyond independent Poisson counts. The Euclid covariance study uses 1000 Euclid-like light-cones from PINOCCHIO and shows that, at Euclid depth and area, the uncertainty budget for cluster number counts must include both shot noise and sample variance induced by large-scale density fluctuations. The total covariance is therefore written as

$12\,\mathrm{deg}^2$0

This paper is not about catalogue construction, but it defines the statistical layer needed for any cosmology-ready ECGC number-count analysis (Fumagalli et al., 2021).

The analytical framework adopted is the Hu & Kravtsov (2003) model. In the Euclid-like mock tests, the analytical covariance reproduces the simulated covariance to within about 10%, which the authors judge sufficient for unbiased Euclid-like parameter inference in this context. Physically, the most important regime is the low-redshift, lower-mass part of the catalogue, where sample variance is comparable to or larger than shot noise; same-redshift correlations between mass bins dominate, but inter-redshift correlations are not negligible. For Euclid requirements, a block-diagonal-only approximation still leads to a potential error-bar underestimate of order $12\,\mathrm{deg}^2$1, so the full covariance is recommended (Fumagalli et al., 2021).

The methodological recommendation is correspondingly specific. The paper concludes that a Gaussian likelihood with cosmology-dependent covariance is the only tested model that yields unbiased parameter inference without underestimated uncertainties. The preferred Euclid-like binning is $12\,\mathrm{deg}^2$2 with $12\,\mathrm{deg}^2$3 mass bins for the Gaussian-covariance treatment. For ECGC users, the implication is straightforward: the catalogue’s cosmological power is inseparable from a validated covariance model, and the abundance likelihood cannot be reduced to independent Poisson bins at Euclid precision (Fumagalli et al., 2021).

This statistical layer also connects back to the catalogue definition. The same paper frames Euclid as a survey expected to deliver a cluster sample of order $12\,\mathrm{deg}^2$4 objects over $12\,\mathrm{deg}^2$5. That scale is precisely why catalogue-level choices—selection function, redshift calibration, and mass-observable modelling—feed directly into covariance-aware cosmological inference rather than remaining local data-reduction details.

6. High-redshift extensions and specialized subcatalogues

The ECGC literature already shows that Euclid cluster products will not be limited to a single all-purpose mature-cluster table. A clear example is the Q1 strong-lensing cluster catalogue, which is explicitly described as an early, specialized Euclid cluster-catalogue product rather than the full ECGC. It was built from Q1 imaging in the three Euclid Deep Fields at Wide Survey depth, but the actual search was conducted on an effective area of about $12\,\mathrm{deg}^2$6 by visually inspecting 1260 preselected cluster fields. The result is 83 unique gravitational lenses with $12\,\mathrm{deg}^2$7, including 14 with $12\,\mathrm{deg}^2$8. The confidence score is defined as

$12\,\mathrm{deg}^2$9

where $1900\,\mathrm{deg}^2$0, $1900\,\mathrm{deg}^2$1, and $1900\,\mathrm{deg}^2$2 denote “Certain Lens,” “Probable Lens,” and “No Lens.” The machine-readable released columns are minimal—name or identifier, RA, Dec, $1900\,\mathrm{deg}^2$3, $1900\,\mathrm{deg}^2$4, $1900\,\mathrm{deg}^2$5, and $1900\,\mathrm{deg}^2$6—which makes this a clean example of a confidence-scored strong-lensing validation table rather than a full cluster meta-catalogue. The paper is explicit that this is a pilot / proof-of-concept / precursor subsample that will also seed automated deep-learning identification of gravitational arcs and multiple images in future Euclid releases (Collaboration et al., 19 Mar 2025).

High-redshift candidate work shows a second branch of the emerging ECGC landscape. The Q1 follow-up of the MaDCoWS2 candidate “Puddle” demonstrates that both Euclid cluster finders can recover a candidate beyond the nominally calibrated regime: the system appears in Q1 as a $1900\,\mathrm{deg}^2$7 overdensity in PZWav with S/N $1900\,\mathrm{deg}^2$8 and as a $1900\,\mathrm{deg}^2$9 overdensity in AMICO with S/N $0.2 \le p \le 1.5$0, while MOSFIRE spectroscopy places the brightest nucleus at $0.2 \le p \le 1.5$1. Within a $0.2 \le p \le 1.5$2 aperture, the Euclid-based overdensity estimate is $0.2 \le p \le 1.5$3 galaxies with $0.2 \le p \le 1.5$4 after completeness correction. The paper explicitly treats this as a Euclid-enabled validation and characterization case study rather than an ECGC catalogue paper, but it demonstrates that joint AMICO+PZWav detection can act as a reliability flag even where Q1 photometric-redshift calibration is still weak (Trudeau et al., 5 Mar 2026).

A broader proto-cluster branch is developed in both observations and simulations. The pilot Q1 study using DETECTIFz searched 11 Euclid MER tiles containing previously known Planck high-star-forming protocluster candidates, covering $0.2 \le p \le 1.5$5. It found 89 raw overdensities with S/N $0.2 \le p \le 1.5$6, retained 10 post-selected protocluster candidates, and focused on 2 that lie within the Planck beams and are independently confirmed by PPM and MC-DTFE-LoG. The same paper gives an empirical forecast of detectability to about $0.2 \le p \le 1.5$7 in the Euclid Wide Survey and $0.2 \le p \le 1.5$8 in the Deep Survey, while emphasizing that projection effects and photometric-redshift ambiguity remain central limitations (Collaboration et al., 27 Mar 2025). In parallel, the Euclid proto-cluster forecast paper treats $0.2 \le p \le 1.5$9 structures as descendant-defined progenitors of $p$0 clusters, shows that their typical angular radii are several arcminutes, and argues that Euclid photometry should be sufficient to build large proto-cluster candidate samples, although spectroscopy is required for confirmation and robust richness characterization (Collaboration et al., 2024).

Taken together, these specialized products indicate that the mature ECGC is likely to coexist with lensing-selected, protocluster, and externally confirmed branches. The published papers do not yet define the final unification scheme for these branches, but they already establish the underlying pattern: Euclid cluster science is being built as a catalogue family in which the main optical cluster list, the external-validation layer, and several science-ready high-purity subsets are structurally distinct but scientifically coupled.